The 2026 Hyperscaler IPO Flood
Earlier this year, I wrote an essay that relaunched The Benchmark.
I argued that the financial world had split in two: The old world — debt, policy, paper — crumbling under its own weight, and the new world — compute, code, energy — ascending.
I made the case that smart money was moving toward the engines driving the new; the hyper-efficient hyperscalers rewriting the DNA of the entire connected world with artificial intelligence.
Companies like OpenAI, Anthropic, Databricks and xAI.
Just these four companies have a combined valuation of about $3.12 trillion — nearly the GDP of France, which ranks seventh in the world.
That’s about 5.4% of the S&P 500.
An average valuation of $624 billion apiece, while those in the index average just $116 billion.
These numbers bring us to the topic of today’s Benchmark:
Most of the companies powering the biggest technological leap since the internet are not listed on the stock market.
At least, not yet.
So what might happen if and when the hyperscaler cohort starts publicly trading?
New > private > public
The stock market exists, in theory, to give the public access to the best businesses in the world.
For most of modern financial history, that contract held.
When railroads transformed the 19th century economy, you could buy railroad stocks. When oil rewired the 20th, you could buy oil majors. When the internet changed everything, you could — eventually — buy Amazon, Google, Apple.
The pattern was pretty much always the same:
Transformative technology. Private phase. Public phase.
Right now, in May 2026, we’re on the precipice of the public phase for the AI hyperscalers. xAI/SpaceX could list as early as July.
Which could mean an historic flood of cash into the stock market.
The four most anticipated IPOs of 2026 represent a combined estimated value of $3.12 trillion.
To put that in context: the entire US IPO market in 2021 — the most active year in a generation — raised around $300 billion.
This pipeline could be ten times that. And it's concentrated almost entirely in AI.
OpenAI is targeting a $1 trillion debut.
SpaceX and xAI, now merged, are eyeing $1.75 trillion combined.
Anthropic — the company whose tools I’m currently using to research and assemble this very newsletter — has a valuation of $380 billion right now, which as I write this could be about to more than double:
Databricks — a less public, enterprise-focused AI business — has crossed a $5.4 billion annual revenue run rate, growing at 65% year-on-year. It is preparing to list in the second half of this year.
All this value is locked in the private market. But potentially not for much longer.
So it’s worth examining some examples of waves of new companies and capital going public.
2026 = 2000?
The standard dot-com cautionary tale goes like this: Zero-revenue companies. Rampant, unchecked speculation. A bubble that was always going to pop, and promptly did, as you can see:
The NASDAQ hit 5,048 on March 10, 2000, then shed 78% in just over two years, hitting 1,114 on October 9, 2002.
So would a wave of AI hyperscaler public listings create a similar event?
First, Pets.com had no revenue.
Anthropic has $14 billion in annualized revenue and growing.
OpenAI is on track for $25 billion.
These aren't dreams dressed up as businesses. These companies are driving a trend, rather than chasing it.
The more important dot-com parallel isn't about the quality of the companies. It's about what a concentrated wave of massive listings does to market structure.
Between 1999 and 2000, hundreds of companies went public in the United States.
The problem wasn't that they were all frauds. It was capital rotation. Every dollar that chased a new tech listing had to come from somewhere.
It came out of existing positions — industrials, financials, consumer staples.
For a time, this looked like genius. Then the rotation reversed, and the unwinding wasn't just a tech correction. It reshuffled the entire market.
In 2000, the companies going public were mostly pre-revenue stories asking markets to fund a dream.
In 2026, the companies preparing to list are already generating billions. The dream has partially arrived.
Which means the capital rotation, when it happens, will likely be validated by earnings, as opposed to narrative — at least initially.
That changes the dynamic. But it doesn't answer the harder question:
Can markets actually absorb what might be coming?
The largest IPO in history (so far)
In December 2019, Saudi Aramco (which has a strange connection to my high school) went public at a $1.7 trillion valuation — the largest IPO in history.
They raised $29 billion, and briefly touched a $2 trillion valuation.
But the stock has been largely flat to down in real terms since then. Current valuation is about $1.8 trillion.
Which means that, what was at the time the world's most profitable company, with government backing, listing on a friendly domestic exchange…
… hasn’t performed amazingly well in the seven years since it listed.
So what might happen given that the four major hyperscalers we’re considering here represent a combined valuation of about twice Saudi Aramco’s?
The mechanics of absorption at scale are underexplored in the current conversation about AI IPOs. Everyone's just debating whether the valuations are real.
Few are asking where the money will come from.
If OpenAI enters public markets at a trillion dollars, index funds will eventually be forced to own it.
Because once a company crosses the threshold for S&P 500 inclusion, every passive fund rebalances. So does every superannuation fund. So does every retirement account tracking the index.
Tesla's inclusion in December 2020 was, at the time, the largest single addition to the S&P 500 in history. Index funds and ETFs bought up about $220 billion. Many passive funds had to do so immediately upon the stock listing.
Tesla entered the S&P 500 at a market cap of around $650 billion, having IPO’d 10 years earlier.
OpenAI is targeting a $1 trillion IPO valuation — roughly 1.5X Tesla's size at inclusion.
But the more striking comparison is the index weight. Tesla entered at about 1.69% of the S&P 500. At $1 trillion, OpenAI would enter at somewhere around 1.7-2% of the current S&P 500 — similar weight, but on a much larger index, meaning the absolute dollar amount of forced buying could dwarf the Tesla IPO.
It’s worth noting, by the way, that Tesla’s valuation today is about 2.5X what it IPO’d at. Its stock price is up about 11X.
But AI hyperscaler IPOs Saudi Aramco and Tesla were not.
Luckily, we already have a relevant AI IPO to look at.
CoreWeave: The hyperscaler IPO guinea pig
CoreWeave’s, er, core business is simple:
It owns massive clusters of Nvidia GPUs, houses them in data centres, and charges AI companies by the hour to use them.
OpenAI is its largest customer — which is both its greatest strength and its most cited risk. One customer representing that much revenue makes investors nervous.
CoreWeave went public on the Nasdaq in March 2025, raising $1.5 billion at a $23 billion valuation. It was the largest US tech IPO in years at the time.
Since then?
The stock has swung violently.
At one point it traded 52% below its post-IPO high.
The markets couldn't decide whether CoreWeave was critical AI infrastructure or an over-leveraged GPU rental business dressed up in a compelling narrative.
You could argue that it's both. And that ambiguity might be a problem for all the AI IPOs in the pipeline.
The order of this IPO queue matters.
If CoreWeave stabilizes and Databricks lists cleanly in the second half of this year, the path could open.
Institutional appetite could build.
The narrative around these businesses could solidify and shed the 2000 comparisons.
OpenAI and Anthropic could enter as headliners to a well warmed-up audience.
But should Databricks stumble — or if the public offering reveals something the private valuation was obscuring — the queue could get very long very fast.
The locked room opens either way. The question is whether it opens onto a welcoming market, or a crowded exit.
History suggests the flood, when it comes, is never as orderly as the companies in the queue would like.
The dot-com wave created a generation of investors who learned — often painfully — how to value internet businesses. This wave will create a generation who learns how to value AI businesses.
How will history look back at the 2026 AI IPOs?
The question nobody can answer yet is whether public markets are ready to absorb the new world's best businesses — or whether the act of absorbing them will reshape markets in ways nobody has fully modelled.
CoreWeave is probably the data point to watch.
Not because it's the most important company in the pipeline. But because it's the first.
How it trades over the next six months could reveal more about what's coming than any analyst note or IPO prospectus.
If the flood comes and markets absorb it cleanly, history might look back at 2026 as the year the new world finally became investable.
If it doesn't — if the weight of more than $3 trillion in private valuations entering public markets triggers the kind of rotation and reshuffling that 1999-2022 did — we'll look back at it differently.
Either way, we are monitoring this situation.
This week's quote:
"Prediction is very difficult, especially about the future."
— Niels Bohr
Invest in knowledge,
Thom
The Benchmark
Share: Forward this email to someone you know would appreciate it.
New here? Subscribe to The Benchmark.
Enjoyed this issue?
Subscribe to The Benchmark
Weekly insights on markets, investing, and portfolio strategy.